TY - GEN
T1 - Online news veracity assessment using emotional weight
AU - Tarmizi, Fatin Amanina Ahmad
AU - Tan, Phan Xuan
AU - Sharif, Khaironi Yatim
AU - Kamioka, Eiji
N1 - Publisher Copyright:
© 2019 Association for Computing Machinery.
PY - 2019
Y1 - 2019
N2 - Trillions of data are being created every day on the Internet due to the growing number of social platforms on the World Wide Web (WWW). Processed data when given in context makes information of any knowledge. However, irresponsible use of the data or misinterpretation of data could be the reasons for false information dissemination. Many researchers from various fields, such as computer science and social science, draw their focus on assessing the veracity of information. There are many techniques to perceive this topic, for instance, social network behaviour, and semantic analysis. The common practice is using semantic analysis approach, where the syntactic structure is analysed and polarity of the texts is determined. In this paper, we approach the veracity assessment by using emotion analysis. We identified emotional states conveyed in news content and calculated the weight of each state in each news content. Contrary to popular belief, our finding showed that emotional, or affective states conveyed in false news are varied - positive and negative states. The distinct feature is the weight of the states in news content. Using multi-layer perceptron, we classified the news and achieved 90% accuracy with our collected dataset and 85% using LIAR dataset.
AB - Trillions of data are being created every day on the Internet due to the growing number of social platforms on the World Wide Web (WWW). Processed data when given in context makes information of any knowledge. However, irresponsible use of the data or misinterpretation of data could be the reasons for false information dissemination. Many researchers from various fields, such as computer science and social science, draw their focus on assessing the veracity of information. There are many techniques to perceive this topic, for instance, social network behaviour, and semantic analysis. The common practice is using semantic analysis approach, where the syntactic structure is analysed and polarity of the texts is determined. In this paper, we approach the veracity assessment by using emotion analysis. We identified emotional states conveyed in news content and calculated the weight of each state in each news content. Contrary to popular belief, our finding showed that emotional, or affective states conveyed in false news are varied - positive and negative states. The distinct feature is the weight of the states in news content. Using multi-layer perceptron, we classified the news and achieved 90% accuracy with our collected dataset and 85% using LIAR dataset.
KW - Affective science
KW - Deception detection
KW - Text analysis, emotion analysis
KW - Veracity assessment
UR - http://www.scopus.com/inward/record.url?scp=85066931239&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85066931239&partnerID=8YFLogxK
U2 - 10.1145/3322645.3322688
DO - 10.1145/3322645.3322688
M3 - Conference contribution
AN - SCOPUS:85066931239
SN - 9781450361033
T3 - ACM International Conference Proceeding Series
SP - 60
EP - 64
BT - ACM International Conference Proceeding Series
PB - Association for Computing Machinery
T2 - 2nd International Conference on Information Science and System, ICISS 2019
Y2 - 16 March 2019 through 19 March 2019
ER -